Correlating Lab Test Results in Clinical Notes with Structured Lab Data: A Case Study in HbA1c and Glucose

Area of Science:

• Clinical Informatics
• Biomedical Natural Language Processing

Background:

• Electronic Health Records (EHRs) primarily structure lab test results.
• Unstructured clinical notes often contain crucial lab data for referral or non-clinical settings.
• Secondary data use for research and practice necessitates robust information extraction methods.

Purpose of the Study:

• To develop and evaluate a rule-based system for extracting lab test results and temporal information from unstructured clinical notes.
• To assess the system's performance in retrieving glucose and HbA1c values.

Main Methods:

• A rule-based information extraction system was designed.
• The system processed unstructured clinical notes from 104 diabetes patients (1996-2015).
• Extracted lab test results were correlated with structured EHR data.

Main Results:

• The system achieved high F1-scores for glucose (0.964) and HbA1c (0.967) extraction.
• Overall extraction performance reached an F1-score of 0.966.
• Successful correlation with structured lab test information was demonstrated.

Conclusions:

• Rule-based NLP effectively extracts critical lab test results from clinical notes.
• This approach enhances the secondary use of clinical data, particularly for temporal lab values.
• The system shows high accuracy for extracting glucose and HbA1c results.